Subject: Re: more work in rc.d [was Re: rc, rc.shutdown proposed change]
To: NetBSD Userlevel Technical Discussion List <tech-userlevel@netbsd.org>
From: Greywolf <greywolf@starwolf.com>
List: tech-userlevel
Date: 03/18/2000 22:33:18
On Sat, 18 Mar 2000, Greg A. Woods wrote:
# Yes but my point was that it still doesn't make any real difference --
# it is just an emotional issue. When it becomes an emotional issue the
# supporters of each particular opposing feature will claim superiority
# whether one's really better than the other or not.
Then you have missed the real point. UI feel is important. I'm a
sysadmin. I do it for a living. UI feel is _real_ important to me.
# I can promise you that if you'd started by learning V7 and 32V, then
# suffered under several years of running Xenix and SysIII, and finally
# been relieved a bit by the last releases of SysVr3, and on occasion had
# suffered with fall backs to SunOS-3 and even SunOS-4, and all the time
# working with at least a few systems that had to work reliably in
# commercial production environments, you'd have a much different view and
# be far more accepting of change in those systems regardless of whether
# it's always for the better! :-)
1. You're a SysV veteran. I'm a BSD veteran. I expect our viewpoints
to be slightly askew each other's.
2. I didn't have your experiences: I was born and raised on four washers
and a fridge (VAX 11/750, three 80MB drives and a TapeStretch-11),
migrated to Sun-2 running SunOS 3, to Sun-3 running SunOS 3, to Sun-3
limping along on SunOS 4, to SPARC box running SunOS 4, to SPARC box
crawling on Solaris, with a few lurid interludes with SVR0, SVR2 and 3
(each with NFS, NIS and a Berkeley TCP/IP stack glued on), and A/UX
(which I did a (n admittedly miniscule) bit of work on all back between
the Sun3/SunOS3 and Sun3/SunOS4, as well as a few random encounters
with IRIX, HP-UX and AIX.
# > If you're expected to learn to admin a system, much less use one,
# > what's the FIRST thing you're gonna see...?
#
# The manual pages.
<dry>
Wow. A system that shows you the manual without intervention from the
minute you log in. Now that's intriguing.
</dry>
# *EVERY* "vendor's" system is different. Trying to
# stuff systems into one of two categories is only going to get you into
# trouble. Everyone stole something from everyone else and everyone made
# their mark with something new and unique. Never ever trust that you can
# guess where something is or how something works -- always check the
# documentation and double check with hands-on tests.
Ain't THAT the trough.
#
# > And I concur with der Mouse: To be told that the details of administrating
# > a system are unimportant is an insult to my career as a systems and network
# > administrator.
#
# That's not waht I was saying.
No, that's what we are being told. "It's all in the subroutines which
you should never need to look at."
Now, I have a bit more TRUST in NetBSD trying to do that, but it's magic
which is definitely on the darker side of gray.
# The point is that narrowing your view of
# how to do things so that you're only competent in one environment is
# limiting.
I will grant you that. You've a very valid point. To my defense,
I am not incompetent -- in fact, I'm reduced to doing even some NT
administration these days! -- but I am a hell of a lot more comfortable
in a BSD environment than not (ps, df, su, and LINES=34 are only a few
of the aggravations I can name).
# Criticising differences just because they are different
# doesn't help. Complaining that "the other" way of doing things is
# harder without learning it openly and using "it" in the environments it
# was designed for doesn't help either.
Nor does approaching it with the attitude of "F*ck you. We're doing it
this way and your opinion as a user doesn't matter. Deal."
It's a problem.
# > Define "production quality". What's _really_ wrong with what we had?
# > "Couldn't be automated." Bull[#@$(]. It could be. Just takes some
# > more thought, is all.
#
# Production quality has to do with things like reliability,
# reproducibility, consistency, integrity, and automatability.
Agreed with everything but the last one. Do you *really* trust any
automation to be done right by _every_ script? Do you really trust
the third parties to have their order in sync with your system?
I am certainly not so foolish.
Now, that aside, I must confess to having installed ssh, gone into
/etc/rc.local, s/pkg/local/g, and rewritten /usr/local/etc/rc.d/ssh
so that it took stop/start args...
# An ideal
# production system would run itself in a "lights-out" 7x24 environment,
# upgrade itself, tune itself, and perform its own disaster recovery.
# All the system administrator would have to do is the initial install, plug
# in new hardware when it's required and authorise the addition and
# deletion of user accounts, software, etc. Obviously no system yet meets
# all of those goals 100%....
Thank you SO much for working so hard to make my job obsolete.
# > Production-quality sysadmin tools != dumbing down the tools.
#
# Of course not -- it means making them smarter, more reliable, and more
# consistent, and more automated.
"Dumbing down the tools" : I meant just the transformation you described.
Hiring a monkey to do systems administration. Thankfully, sysadmin is
more than just that.
# > Your first production-quality sysadmin tool had better be your systems
# > administrator!
#
# Well, hmm... from a systems programmer's P.O.V. I would have to
# disagree! :-)
Of course. I'm a systems administrator (and I do my job very well,
thanks).
I'm a systems administrator because, admittedly, I can script and I
can program my own tools as necessary. I've never been put on a
programming project from start to finish (and I can't do that on
my own, really.).
# Greg A. Woods
#
# +1 416 218-0098 VE3TCP <gwoods@acm.org> <robohack!woods>
# Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>
#
--*greywolf;
--
BSD: The Power of Code.